nguyenkhoa
commited on
Commit
•
5c7926a
1
Parent(s):
ff1f075
Upload tokenizer
Browse files- tokenizer_config.json +4 -0
tokenizer_config.json
CHANGED
@@ -48,10 +48,14 @@
|
|
48 |
"eos_token": "</s>",
|
49 |
"errors": "replace",
|
50 |
"mask_token": "<mask>",
|
|
|
51 |
"model_max_length": 16384,
|
52 |
"pad_token": "<pad>",
|
53 |
"sep_token": "</s>",
|
|
|
54 |
"tokenizer_class": "LEDTokenizer",
|
55 |
"trim_offsets": true,
|
|
|
|
|
56 |
"unk_token": "<unk>"
|
57 |
}
|
|
|
48 |
"eos_token": "</s>",
|
49 |
"errors": "replace",
|
50 |
"mask_token": "<mask>",
|
51 |
+
"max_length": 4096,
|
52 |
"model_max_length": 16384,
|
53 |
"pad_token": "<pad>",
|
54 |
"sep_token": "</s>",
|
55 |
+
"stride": 0,
|
56 |
"tokenizer_class": "LEDTokenizer",
|
57 |
"trim_offsets": true,
|
58 |
+
"truncation_side": "right",
|
59 |
+
"truncation_strategy": "longest_first",
|
60 |
"unk_token": "<unk>"
|
61 |
}
|